| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | REGISTER |
|
|
|
|
||||||||||||
|
|
||||||||||||||
To:
ADC Online Letters and ADC Education and Practice Letters
Electronic Letters to:
|
|
Electronic letters published:
![[Read eLetter]](/icons/misc/sectionDOWN.gif){kind=icon}
Pooling of trials is not appropriate in case of heterogeneity
|
|
|||
|
Casper W Bollen, Pediatric Intensivist University Medical Centre Utrecht, Cuno SPM Uiterwaal, Adrianus J van Vught
Send letter to journal:
c.w.bollen{at}umcutrecht.nl Casper W Bollen, et al.
|
Dear Editor We read the systematic review by Thome et al. with great interest on the elective use of high frequency ventilation compared with conventional mechanical ventilation in preterm infants.[1] Thome et al. included 17 randomized trials and stated that unlike previous meta-analyses they did not find significant benefits in pulmonary outcome. They also referred to our published cumulative meta-analysis.[2] However, we would like to point to the fact that in our meta-analysis we reported the same finding, but we restricted pooling to subgroups according to ventilation strategies used for both high frequency ventilation and conventional mechanical ventilation. The reason we did not pool all studies was because there was considerable heterogeneity between studies. The use of random effect models to overcome the problem of heterogeneity is debatable.[3] Thome et al. showed an odds ratio of 0.87 with a 95% confidence interval of 0.75 to 1.00. Strictly speaking, the effect of high frequency ventilation compared with conventional mechanical ventilation was not statistically significant. However, these results would leave ample room for debate whether or not high frequency ventilation confers favorable pulmonary outcome. Instead, comparing high frequency ventilation with a high lung volume strategy with optimal conventional mechanical ventilation with a lung protective strategy resulted in our study in a relative risk of 0.95 with a 95% confidence interval of 0.85 to 1.07. Thome et al. used the same technique of recursive meta-analysis as we used in our publication to quantify the relative change in effect size between trials over time and attributed these changes to differences in ventilation strategies used in trials. To establish a causal relation between differences between trials and reported effect estimates is problematic. Such an analysis would be observational with trial as unit of measurement. As in all observational research, associations could be confounded by covariates. For example, the fact that Thome et al. did not find an association between time lag of enrolment and outcome could be confounded by the covariate ventilation strategy used in conventional mechanical ventilation. In conclusion, we fully agree with Thome et al. that ventilation strategies used in both high frequency ventilation and in conventional mechanical ventilation seemed to play an important role in determining outcome. However, pooling of all studies in spite off considerable heterogeneity was not appropriate. The exact causal relationships of differences between trials in ventilation strategies, time of enrollment and population characteristics with effect sizes have yet to be established. References 1. Thome UH, Carlo WA, Pohlandt F: Ventilation strategies and outcome in randomized trials of high frequency ventilation. Arch Dis Child Fetal Neonatal Ed 2005. 2. Bollen CW, Uiterwaal CS, van Vught AJ: Cumulative metaanalysis of high-frequency versus conventional ventilation in premature neonates. Am J Respir Crit Care Med 2003, 168: 1150-1155. 3. Egger M, Smith G, Altman D: Systematic Reviews in Health Care, 2nd edn. Blackwell BMJ Books; 2001. |
|||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | REGISTER |
| ARCH DIS CHILD | FETAL NEONATAL ED | ED PRACTICE |